3

Zeno and the Mathematicians

At some time in the first half of the fifth century BC, Zeno invented the set of 143
paradoxes on which his successors have sharpened their wits. The puzzles
have come down to us in various versions, more or less incomplete and more
or less reflecting the special interests of later writers. What we have left of
Zeno's best-known work comes, on the most hopeful view, to less than two
hundred words. Still, this has not stopped Zeno's admirers from trying, with
all due caution, to reconstruct the programme of all or some of his arguments.
I want to make one such programme plausible and to show how, if I am right,
this makes some solutions to the puzzles beside the point.
My second interest in the paper is this. Zeno, it is commonly said, was and
wished to be the benefactor of Greek mathematics. By his day the
Pythagoreans had brought mathematics to a high level of sophistication. But
the foundations of their system were a nest of confusions, and Zeno was out
to expose these confusions. One beneficial result of his arguments (on this
familiar account) was to compel mathematicians to distinguish arithmetic
from geometry.
This picture seems to me mistaken. Zeno neither had nor tried to have this
effect on mathematics (though in other ways, no doubt, he did influence
contemporary work in the science). But his arguments had a great effect on a
later stage of mathematics, and the effect was not beneficial./
{.eno 's programme 144
Zeno certainly held, as a philosophical theory inherited from Parmenides,
that there is only one thing in existence. This is an embarrassment to those
who want to portray him as trying to set up a consistent logic for analysing the
structure of space and time. For it means that he thought there was no such
structure: any way of dividing things in time or space must carry absurdities.
If this was his theory we should expect him to work out an exhaustive list of
possible ways of dividing things and to set about refuting all the possibilities
separately; and this, as I shall try to show, is what he does.
(Let me say at once that this talk of dividing is deliberately ambiguous. It
is not always clear, for instance, whether Zeno is discussing the possibility of
[Postscript!97!: I have made small corrections to the text as printed in 1958, and marked
later comments by square brackets in the notes.]
46 /. The Presocratics
producing a plurality by actually carving a thing up or by enumerating the
fractions it must logically contain; but for most of the way we shall find the
distinction irrelevant. What matters is that whichever operation Zeno has in
mind he is canvassing its logical and not its physical possibility.)
Some hold that Zeno was not committed to any philosophical tenet
whatever. For he is credited with saying 'Show me what the one is and then I
can tell you what things (in the plural) are';2 and this is sometimes taken to
show that he did not profess to understand even the one thing that
Parmenides had left in existence. But the point of his words is just that, if you
want to say that there are a number of things in existence, you have to specify
what sort of thing counts as a unit in the plurality.
3
If there can be no such
individuals as you claim there can be no such plurality either. And in
particular if your individuals have to be marked off by spatial and temporal
distinctions you have to be sure that your way of making such distinctions is
not logically absurd.
Plato makes it clear that Zeno's major work was divided into separate
arguments, each depending on some hypothesis and reducing the hypothesis
to absurdity. We do not know the content of these hypotheses, but Plato is
emphatic that every argument was designed to refute the proposition that
there are a number of things in existence. We certainly have reports of some
of the arguments which began 'Suppose many things exist.' But we also have
145 a report of one which starts 'Suppose place exists. '
5
/ And Aristotle treats the
familiar puzzles of Achilles, the Arrow, the Stadium and the Dichotomy as though
these were designed in the first instance to refute the possibility of movement,
not of plurality. It might be, of course, that these latter arguments came from
another work of Zeno's.
6
But I shall try to show that they play an essential
part in the attack on plurality.
1
It is not clear even that Zeno used the word diairtin and its cognates; but Parmenides had,
and Zeno certainly used equivalent language in discussing Parmenides' topic (see A(l) below).
2
Eudemus apud Simpl. in Phys. 97. 12-13, 138.32-3.
'Cf. Alexanderapud Simpl. in Phys. 99. 12-16.
Parmtnidts 128e-128a, a version which became standard with later commentators (e.g.
Simplicius in Phys. 139. 5-7.
' [Or so Diels-Kranz (i 498. 8-10) infer from Aristotle, Phys. 210b23, and its repetition by
Simplicius (562. 4). But it seems likely from Aristotle's first statement of the puzzle at 209a24-5
and Simplicius' note there (534. 9-11), as well as from Eudemus (apud Simpl. 563. 18-20; cf.
Philop. 510. 2-4, 599. 1-3 tl al.), that the hypothesis was 'If each existing thing is somewhere' (or
perhaps 'in (a) place' or 'in something', but cf. Simp I. 534. 10, 563. 20-4). And then the connection
with plurality is evident. On the ordinary assumptions of physical divisibility that we shall see
Zeno exploiting, plurality calls for spatial distinctions. So the force of Zeno's puzzle may have
been: if each thing must be marked off by being somewhere, in some place, then that place in turn
is a thing to be marked off, ex hypothesi, by being in some place; and so ad inf (That the argument
was a regress is clear from Aristotle, II. cc.) There is no basis from which to start in locating
things that are spatially distinct, for locations are themselves such things.)
. ' Not that the evidence that he wrote other works is strong: Plato seems not to know of them,
yet he certainly knew of the arguments on motion (cf. Phatdrus 261d and the application of the
Arrow in Parmtnidts 152b-e). [Of the four titles credited to Zeno by the Suda (29 A 2 DK) three
are readily assigned, as librarians' variants, to the book known to Plato. The fourth, 'Exegesis of
Empedocles' Verses' (Diets' evidence that the title could signify a polemic dissolves on scrutiny),
3. <:,eno and the Mathematicians 47
Zeno's major question then is: if you say there are many things in existence
how do you distinguish your individuals? The answer in which he is chiefly
interested is that the world and any part of it can be broken down into its
individual parts by spatial and temporal divisions. And the paradoxes that I
am anxious to discuss are those designed to meet this answer, namely those
which are jointly planned to show that no method of dividing anything into
spatial or temporal parts can be described without absurdity.
For suppose we ask whether such a division could be (theoretically, at least)
continued indefinitely: whether any division can be followed by a
sub-division, and so on, through an infinite number of steps. Let us say, to;
begin with, (A) that it does have an infinite number of steps. Then could such
a division nevertheless ever be (or ever have been) completed? (A 1) One of
Zeno's arguments is designed to show that it could not.
The paradox had two arms. The first began by arguing that the units in a
collection can have no size at all: else they would have parts and be not units
but collections of units.
7
The second began by arguing that, on the contrary,
there cannot be anything that has no size at all; for there cannot be a thing
which if it were added to or subtracted from something else would not affect
the size of that thing.
8
So the first arm of the argument assumes that the units
it describes are theoretically indivisible; and the point of this requirement
comes out in the sequel, when Zeno shows that he is discussing the class of
individuals produced by an exhaustive division of something, a division whose
end-products cannot themselves be further divided. The second arm of the
argument assumes, on the other hand, that its units must be capable of being
added and subtracted in a sense in which these operations cannot apply to
things without magnitude; and the point of this requirement comes out in the
same sequel, for if a thing can be divided into parts (exhaustively or not)
those parts/ must be capable of being added to make the thing, and in that 146
case they must have some size, however small.
Next, to bring these requirements into one focus, Zeno went on to specify
the collection of parts in which he was interested, namely the collection
produced by completing a division in which every step has a successor. 'Each
thing', he said, 'must have some size and thickness, and one part of it must be
was quite probably transferred to Zeno of Elea from Zcno of Citium, perhaps through the
tendency of the Stoics to call their founder 'the first Zeno' (Clem. Strom. v 9; there was a later
Stoic Zcno, Diog. Laert, vii 35 ). Heraclitus and Empedocles were of almost equal importance for
the Stoics: between Heraclides' four books on Heraclitus and Cleanthes' four on the same
subject it is unlikely that the Stoic Zcno took no hand in the exegesis (cf. Diels, Doxograplli Graeci
(Berlin 1879), 469-70); and as for Empedocles, Chrysippus introduces some major arguments by
an exegesis of his verses (Galen in SVF ii 237.22-5). The theory of the four elements was
acknowledged to him (Dox. Gr. 92 b-93 b), and by Stoic oiktiosis he was credited with a
cosmology based on final conflagration (Hipp.olytus, RH i 3) and his sphere-god became a noeron
pur comparable to thep/lronimonpurofHeraclitus (ibid. ix 10).]
'Simplicius, op. cit., 139. 18-19: this argument at the start of the paradox is still overlooked by
English editors, although its text and sense were settled by Hermann Fraenkel in AJP, 1942,
14-17 (= WegeundFormmfriihgriechischenDenkms'Munich 1968),211-14.
1
Simplicius, op. cit., 139. 9-15.
48 I. The Presocratics
separate [or perhaps just 'distinct'] from another. And the same holds good of
the part which is in the lead - that too will have some size, and of it too some
part will be in the lead. In fact to say this once is as good as saying it for ever,
for no such part of the thing will be the last or unrelated to a further part. '
9
These words define a division so that there can be no last move in the
sequence: for any fraction that is taken, a similar fraction can be taken of the
remainder (the 'part in the lead'). In this, certainly, there is no clear
implication that such a division can have been completed. But Zeno does
make that assumption in drawing his conclusions. For he points out that, on
one line of argument (that of the first arm), the parts produced by this division
can have no size at all: they are end-products whose further division is
logically impossible. And he also points out that, on the other line of
argument (that of the second arm), since all the parts of such a collection
must have some size the whole collection (and by the same token any part of
it) must be infinite in size. And both conclusions are absurd. They were
presented as an antinomy; but as a dilemma they are equally lethal. Either
the parts have no size, and then there can be no such parts; or they have some
size, and then the thing you set out to divide becomes infinitely big.
10
Notice that Zeno is not first setting up a division which cannot have a last
move and then asking, improperly, what the last move would be.
11
He is
asking, legitimately, what the total outcome of the division would be; and for
there to be such an outcome there must be a smallest part or parts.
The effect of the argument is to show an absurdity in the alternative for
which we opted first, namely that if anything is infinitely divisible such a
division can be carried right through. So now (A 2) we shall say that
anything is infinitely divisible but that such divisions can never be completed.
Then, supposing that the puzzle about Achilles and the tortoise
12
is a puzzle
147 about infinite divisibility, it is designed to block this escape-route. In/ order
to overtake the tortoise Achilles must first reach the tortoise's starting-point;
but by then the tortoise will have reached some further point. So then Achilles
'Simplicius, op. cit., 141. 1-6. A commoner but linguistically less easy version of the words
runs 'Each thing must have some size and thickness and there must be another thing separate
from it. And the same holds good of the thing in front: it too will have some size and there will be
something in front of it ... 'Taken in this way the words do not define the steps in the division but
merely characterize its products by saying that the series has no last member. And there is no
mention of pans (more exactly, none of the Greek genitives is understood as panitive) before the
last line. Otherwise, for our purposes both versions come to the same.
10
[That Zeno is here assuming, per impossible, that the division is completed is common doctrine
to the Greek commentators: cf. Porphyry apud Simpl. in Pltys. 139. 26-40. 6 (Simplicius rightly
corrects Porphyry's slip and follows Alexander in assigning the argument to Zeno, 140. 21-6).
Porphyry's account of the argument, and panicularly the explanation in 140. 1-5, notoriously
echoes Aristotle's discussion in GC 316a14-34 (see D.J. Furley, Two Studies in the Greek Atomists
(Princeton 1967), 84-5), and in panicular his diJiristhO pantii (140. 2) recalls Aristotle, 316a23-4;
but without begging the question this cannot be argued to show that the gist of the argument
. was the invention of Aristotle and not, as the commentators claim (and as I have tried to argue
from the texts they repon), ofthe Eleatics. It is not relevant to criticize Zeno's arguments without
noticing this assu.mption.]
11
Cf.J.F. Thomson, 'Tasks and Super-tasks', AMiysis 15 (1954-5), 6-7.
12
Aristotle, Physics 239b 14-29.
3. <.eno and the Mathematicians
49
must reach this point, by which time the tortoise will have got on to another,
and so forth: the series comes to no end. The moves which Achilles is
required to make correspond to divisions of the intervening country, and the
divisions are infinite, determined by the same general formula as in A 1. But
on our present assumption Achilles cannot complete any such sequence of
moves; so he cannot overtake the tortoise, whatever their relative speeds and
however short the lead.
Now if I am right about the coupling of Zeno's arguments
13
it is beside the
point to maintain, as a general solution of this puzzle, that an infinite division
can be completed. For if we say this Zeno will take us back to A t and ask us
about the character of the parts produced by the division. To make
this clear, consider Aristotle's first solution to the puzzle- a solution which he
later admits to be unsatisfactory but which he nevertheless thinks to be
adequate ad hominem.
14
He replies that, provided we recognize that the time of
the run can be divided in just the same way as the ground, Achilles can
overtake the tortoise in a finite time; for the smaller his moves become the less
time he needs to accomplish them, and these component times can diminish
without limit. Then suppose we tdl Achilles to mark in some way the end of
each stage of the course in which he arrives at a point reached by the tortoise
in the previous stage. Suppose also we satisfy Aristotle's requirement and
allow these successive markings to follow each other at a speed which
increases indefinitely, in inverse ratio to the ground covered at each stage;
and suppose the marks become proportionately thinner and thinner. Then
Zeno, as I understand him, argues that if Achilles claims to have finished his
task we can ask about the positions of these marks, and in particular of the
last two. If they are in the same place there is no stage determined by them,
and if there is any distance between them, however small, this distance is the
smallest stage in an infinite set of diminishing stages and therefore the course
is infinitely long and not just infinitely divisible.
Two things, I take it, we must give Zeno: first, that of the series of
movements that Achilles is supposed to make there can be no last member,
just as of the stages of the division described/ in A t there can be no last 148
stage; and second, that if either series can be completed it must be possible to
describe the resulting state of affairs without absurdity. From these
admissions Zeno infers that Achilles can never finish the run that brings him
level with the tortoise. Any hope of salvation lies in looking at this inference.
Consider that other series of moves to which Professor Black once likened
Achilles' run.
15
Hercules is required to cut off the Hydra's heads, but every
time he cuts off a head another grows in its place. When can he finish the
assignment? Never, if the task was correctly specified. For if some heads are
left on Hercules has more work to do, and if all are off it was not the case that
"Notice the proecluin which may have been common to both puzzles: Simplicius op. cit., 141.
4; Aristotle, op. cit., 239b 17.
14
op. cit., 233a21-31, 263at5-18: the solution is applied first to the Dichotomy, discussed below,
but Aristotle took this to be the same puzzle as the Achillts.
"M. Black, 'Achilles and the Tonoise', Analysis 11(1950-51), 98 (= Prohltms of Analysis
(Ithaca 1954 ), I 05 ).
50 /. The Presocratics
for every head cut off another head grows. But these are exhaustive
alternatives, so there is no subsequent state of affairs of which it is logically
possible to say truly: Hercules has finished his task. Now (as Mr Watling has
already argued)
16
this is not the case with Achilles. There are plenty of states
of affairs compatible with Achilles' having achieved his task of overtaking the
tortoise: plenty of positions beside or beyond the tortoise that Achilles can
have reached. It is just the case here that Achilles' movements have been so
described that they have no last term, but not so that no subsequent state of
affairs is compatible with his having completed the series. But to require
anyone to finish an infinite division, as in A 1, is to start them on a
Hydra-operation: there can be no state of affairs, no collection of bits, of
which it is possible to say: Now the job is done. For either the bits do or they
do not have some size, and that exhausts the subsequent possibilities. On this
Zeno was right. His error was to construe his A 2 example on the model of
Al.
In a later paper
17
Black admits this difference in the sense in which
Hercules and Achilles can be said to have taken on an infinite set of tasks.
But he still holds that in either case 'talk of an infinite series of acts performed
in a finite time is illegitimate'. For he now says that the description of
Achilles' movements belongs to 'common-sense language' which, in contrast
to the mathematical representation of space and time, 'does not permit talk of
the indefinitely small' - that is, does not have a use for describing Achilles'
movements as becoming as short as you like. But a guillotine is not an
argument. If someone says, 'In making any movement you make an infinite
149 series of decreasing/ movements', we have no reason yet to reject this as an
offence to common usage. It already looks like a recognizable application of
mathematical language to the description of familiar events (we recall the
graphic problems in school arithmetic): what it needs at once is clarification.
We can ask 'What do you mean here by "infinite series"? Do you say that in
walking from a to d I make a set of smaller walks of which the first takes me
beyond a and the last brings me level with d? For then I cannot see how you
define this sequence so as to let me draw on my knowledge of other uses of
"infinite".' Suppose then he gives us a formula, as in A 1 or A 2, for defining
the class of movements so that there can be no last move in the sequence
bringing me level with d. Then we know how he is using the redescription of
our movements that he has introduced. He has not uncovered an unsuspected
set of events in our daily histories and he has not burdened us or Achilles
with a new and crippling set of duties: the connection between our usual
descriptions of Achilles' run and this sort of restatement is not in either of
these ways a factual connection. What we have been given is a translation of
those usual descriptions; where the second can be known, directly, to apply,
the first can be known, derivatively, to apply. So no consequential question
can arise about the applicability of the second. If we are told that the equation
shows why Achilles never can catch the tortoise, we can only complain that
"J. Watling, 'The Sum of an Infinite Series', AMiysis 13(1952-3), 41-2
" 'Is Achilles still running?' Problems of AMiysis, I 09-26.
3. -?_eno and the Mathematicians 51
the proffered rules of translation have broken down and go back to our
request for clarification. Any attempt at this stage to reconstrue the
expression 'series of moves with no last members' as specifying a
Hydra-operation, an infinite parcelling of the ground such that no state of
affairs is compatible with its completion, cancels the equation with our
description of Achilles' run. And in this way 'common-sense language' is
safeguarded; for it is the oscillation between bringing in the infinite series as
a logically innocuous translation of ordinary statements and trying to
reconstrue it on the model of the task in A 1 that breeds the puzzle.
18
A closely associated paradox is the Dichotomy.
19
Before reaching your
destination you must reach halfway, but before reaching that you must reach
halfway to it; and so back. So in this series there is no first move, and you
cannot get started. {It can of course also be made to show that there is no last
move; Aristotle seems to take it so. But this was taken care of by the
Achilles.)/ Here again if you insist that there is a first move you are taken back 150
to A 1 : either this move is no move, or it covers some distance, however small.
The solution here is the same as for the Achilles. But Aristotle says that in face
of this puzzle some theorists (certainly Xenocrates and apparently at one time
Plato) postulated atomic distances, 'indivisible lines'.
20
That is, they
challenged Zeno's disjunction 'Either no size at all; or some size, and then
divisible' by adding 'Or some size, but not divisible.' Then the first or last
move towards one's destination would be to cover such an atomic distance;
for one could not logically be required to cover any fraction of it first. It is not
certain whether the proponents of this theory thought that any measurable
distance contained a finite or an infinite number of such distances. An
argument for thinking that they meant the former is that this is assumed in
the fourth-century polemic On Indivisible Lines. An argument for thinking the
contrary is that the theory was held at a time when the difficulties of
incommensurable lines were fully realized. It was a commonplace that the
side and diagonal of a square cannot both be finite multiples of any unit of
length whatever. If the latter account is true, those who introduced this
theory were suggesting that an infinite division can have a last term: the
products of such a division are not completely without magnitude, yet they
11
['Redescription', 'restatement', 'translation' and 'equation' are in fact all terms stronger
than my argument needed. All that is required is that 'moving from A to B' should be
represented as innocuously entailing 'moving to the halfway point between A and B, and the
halfway point of the remainder, and thereafter every such halfway point in this infinite series' (or,
in the case of the Achilles, a more sophisticated variant of this in which the position of the points is
determined by the tonoise's initial lead and speed).The converse entailment is not needed for my
purposes; and Benacerraf challenged it effectively ('Tasks, super-tasks and the modern Eleatics',
]oumtJl of Philosophy 59(1962), 765fT.) by inventing a genie who set out to go from A to B but was
afflicted with Proponional Dwindling. Only a half of the genie was left when he reached halfway,
only a quaner when he was a quaner-way from home, and so on. There would be some of him to
reach ttJtl'} halfway point in the infinite _!leries yet none to arrive at B. Perhaps, in Professor
Wiggins' words, 'if it takes a magical example to prove a conclusion this is significant in itself';
but it need not be punued here.]
"Aristotle, Playsi&s Vl239bll-14 (cf. 233a21-3).
20
Playsit:s 1187a1-3.
52 /. The Presocratics
have no finite magnitude such that fractions of it can be specified. They
would be, in fact, to all present intents and purposes, infinitesimals,
vanishingly small quantities; and movement over such a distance is what
writers on mechanics such as Heinrich Hertz have called infinitely small or
minimum displacements. But whichever interpretation of the theory we give, it
was an attempt to evade Zeno's dilemma in A 1.
Now it looks as though this attempt is met in advance by another of Zeno 's
arguments, that known as the Stadium.
21
On the prevalent interpretation of
the argument this is certainly so; and I wish I could be sure of the truth of the
interpretation. But it is fair to warn you that, if the moral of the argument is
anything like that now found in it, the Greeks seem to have missed the point
by a wide margin. Plato, who converted many of Zeno's arguments to his
own use, made no use of this one and apparently saw no objection to
postulating infinitesimals. Aristotle rejected infinitesimals, but he missed the
sense of an argument that Plato had missed before him./
151 The puzzle sets up three parallel rows of bodies. All the bodies are equal in
size; each row contains an equal number of them; and (a stipulation omitted
in Aristotle's report) the bodies in each row are directly adjacent. One row
(the As) is stationary. The other two (Bs and Cs) meet at the mid-point of the
As and move on past each other at equal speeds, so that when the first B clears
the last A in one direction the first C, moving in the opposite direction, clears
the last A at the other end. Thus in the time that the first B passes half the As,
from mid-point to end, it passes all the Cs. Let this time be t. But then if the
first B takes t to pass n bodies (to wit, half the As) it must take not t but 2t to
pass 2n bodies ( v i ~ . all the Cs). So the move which takes t also takes 2t; this is
the alleged puzzle, and plainly it depends on disregarding the relative motions
of the bodies. The Cs are moving, the As are not. That is Aristotle's sole
comment on the argument, and it is generally felt that if it is refuted by such
a comment it was not worth the considerable space he gave it.
Suppose now that Zeno asks how we can specify the relative motions of the
bodies. If we say that the first B can pass twice as many Cs as As in a given
time, what we say entails that if in a given time it passes one Cit also passes
half an A. But suppose now that any A (and therefore any B or C) is an
infinitesimal quantity. Then the B cannot pass half an A: it must pass all or
nothing. And since ex hypothesi it is moving past the As it must pass a whole A
in the time that it passes one C. Yet, as we set up the problem, it would pass
twice as many Cs as As in a given time. So when it passes one Cit also passes
two Cs, and this gives Zeno his contradiction. It seems the simplest hypothesis
that gives the problem any weight whatever.
22
There is a familiar argument to show that, if lengths are made up of
21
Aristotle, Pllysics VI 239b33-240a 18.
zz But it is possible that Zeno was out to explode the distinction between mouing and static.
Given that the distinction is relative, any one of the rows of bodies could be taken as providing the
units of distance for assessing the speeds of the others. Trading on the fact that no row had prime
right to this status, Zeno gave it to two of the rows in the same argument. [D.J. Furley propounds
a very acceptable version ofthe paradox (Two Studies, 75); it seems to me however to yield a moral
about indivisibility of a kind that he would like to avoid.)
3. ,(,eno and the Mathematicians 53
infinitesimal lengths, everything that moves must move at the same speed.
23
Zeno goes one better than this. He argues (on the present interpretation at
least) that, if bodies are made up of infinitesimal lengths, then even if bodies
do move at the same speed they cannot move in opposite directions.
This argument, then, seems designed to destroy the last hope that the sort
of division described in A 1 could theoretically by terminated, in the sense of
producing any specifiable end-products. And the Achilles. and the Dichotomy
were devised to eliminate the alternative, that the world or any part of it was
open to an in-/finite division that did not terminate in any end-product. The 152
next question is whether Zeno faced the alternative (B) that any division
terminates in some finite number of steps beyond which no further step is even
logically possible.
Against one arm ofthis option he did not, as far as we know, think it worth
arguing, namely the joint assertion that (a) anything is divisible for only a
finite number of steps and (b) the products of such a division will have some
finite size. For this could only be a thesis about physical possibilities: Zeno
assumes without argument in A 1 that the conjunction of size with theoretical
indivisibility would be a contradiction. Suppose, on the other hand, that the
products of such a division are said to have no size: then the argument of A 1
that all parts must have some magnitude goes home against this thesis too.
And suppose it is said that the products are vanishingly small, then the Stadium
argument is equally effective here. For neither of these arguments requires
that" the end-products with which it deals should be produced by an infinite
rather than a finite number of divisions.
However, I think that another of Zeno's arguments may be levelled
directly against option B.
24
This is the argument that a collection containing
a finite number of parts must also contain an infinite number of them. It must
contain just the number that it does, whatever that number is; but between
any two members there must be another member, so that the collection is
infinitely numerous. The writer who reports this argument takes it to be
concerned once again with the results of an infinite division.
25
But it can be
understood more generally, as a foretaste of Bradley's paradox. Any two
members of a collection must be separated by something if they are to be two
things and not one; but by the same argument what separates them must
itself be separated from each by something else; and so forth. I suspect that
this is the correct interpretation because the argument then becomes
complementary to one of Parmenides'. Parmenides had urged that if two
things are separated it must be a gap, nothing; but this is to mistreat nothing
as a substantial part of the world.
26
Zeno reinforces this by extracting a
different embarrassment from the plea that things are separated not by
nothing but by other intervening things, substantial parts of the world. And
his argument begins from the consideration of a finite collection; so it may
" Cf. Russell, Principles of M alhematics' (New York 1943 ), 322.
"Simplicius, op. cit., 140. 28-33.
" ibid. 140. 34-5: Simplicius on his own authority?
26
Diels-Kranz, Vorsokratiker' 28 B 8: 22, 46.
54 /. The Presocratics
153 well/ be aimed at any who thought that there must be some finite number n
such that the world could be divided i.nto n things but not - logically not -
into any number higher than n.
Certainly, this argument seems patently fallacious. For surely things may
be separated by their common boundaries- by their edges, and nothing else.
And it is absurd to ask what separates them from their edges, absurd for the
reason that Plato and Aristotle drove home, that the edge of a thing is not
another thing of the same type as what it borders, not a part that can be cut
off its possessor. The moment that begins a stretch of time or the point that
bounds a line is not any stretch, however small, of time or space. Otherwise it
in turn has a beginning, and then Zeno's regress is afoot. And Zeno is
accused of ignoring this distinction.
If that is so we can turn to the argument through which, if through any,
Zeno exercised a major influence on the mathematics of science. For it is in
this argument above all that he is accused of confusing edges with the things
they border, or more precisely of confusing instants, which are the limits of
time-stretches, with time-stretches. But it seems equally likely that he is now
characteristically trying to seal off an escape-route from the last argument by
showing how absurdities came from the attempt to distinguish moments from
periods of time. This remaining puzzle is that known as the Flying Arrow. But
before discussing it let me bring the mathematicians into the picture.
The mathematicians
Most handbooks written since the time of Paul Tannery will tell you the
purpose of the arguments we have examined so far. By Zeno's day Greek
mathematics, in the hands of the Pythagoreans, had come to exhibit the
familiar picture of a sophisticated superstructure built on badly confused
foundations. In his arguments on divisibility Zeno was out to expose these
radical confusions, and he succeeded.
Following some other writers I am inclined to think this explanation a
myth, and an obstructive . myth. For first, the picture of Pythagorean
mathematics, to the extent that it is intelligible, rests on quite inadequate
evidence. And secondly (and for the present paper more relevantly), if there
154 were such a stage in the/ history of mathematics, Zeno's arguments would
not be directed primarily at it.
Briefly, the theory ascribed to Zeno's contemporaries is this. It is mainly
the work of Paul Tannery, but later writers have added to it. Cornford, one of
the most important of these, credits the Pythagoreans with failing to
distinguish physical bodies from geometrical solids, and with holding about
these solids both that they are infinitely divisible and that they are divisible
into atomic bits, which bits both have magnitude and have the properties of
points without magnitude.
27
Indeed they seem to have held every possible
27
P. Tannery, Pour l'histoire de/a science helltne (Paris 1877), ch. 10; F.M. Cornford: Plato and
Parmemdes (London, 1939), 58-9, and papers in CQ, 1922-3.
3. ,Zeno and the Mathematicians
55
opinion about the divisibility of bodies save the opinion that bodies are not
divisible. Certainly, Zeno was anxious to find confusions in the claim that
bodies are divisible at all. But to ensure that he was writing with a special
target in view the target has been enlarged to the point where a shot in any
direction will hit it.
This is not the place to hold an autopsy on the evidence for this theory.
Much of the work has been done in print,
21
and what needs. to be added can
be deferred. What is to our purpose is that Zeno's arguments cannot have
been directed against such a theory unless his whole programme was
misconceived. For in order to provide his arguments with a target a theory
had to be produced which houses every or nearly every incompatible view on
the divisibility of bodies. But the direct refutation of such a theory would be
to show the absurdity of holding any two or more of these views concurrently.
What Zeno does is to distinguish each view and refute it in isolation. Thus he
deals separately with absurdities arising from the addition of magnitudes (in
A 1 ), although for Tannery the basic confusion in Pythagoreanism was the
confusion between numbers and magnitudes. And he wrings separate
embarrassments from the option that the ultimate parts of things have no
magnitude and the alternative option that they have some magnitude, and
again from the possibilities that a continuous dichotomy can and cannot be
completed. In brief, his arguments seem designed to close not some but all
avenues of escape to anyone holding the unremarkable belief that there is
more than one thing in existence. To suppose that he is merely attacking the
possibility of taking more than one of these avenues at once is to wreck the
structure of his/ arguments and to neglect such evidence, internal and 155
external, as we have of their motivation.
Now let me reset the scene by reminding you of some real teething-troubles
that had overtaken mathematics by the time of Plato and Aristotle. The early
Pythagoreans had certainly worked on the assumption that any two lengths
can be represented as related to each other by a ratio of whole numbers. Any
geometrical theorem could be applied in terms of the theory of numerical
proportion that they had developed on this basis. But before Plato's day this
assumption had run up against the discovery that lines could be constructed
which bore no such proportion to each other. No matter what positive integer
is assigned to the side of a square, no corresponding integer can be found to
represent its diagonal.
29
Some text-books would let you suppose that this discovery compelled
mathematicians to jettison the old theory of proportion. But several reactions
to it were possible. One was to retain the theory but restrict its scope: and this
is just what Euclid does with it in the seventh Book of his Elements. One was
to retain it and apply it to the sides and diagonals of squares by an
accommodation that could be made as small as you please.
30
And one was
21
In particular by Calogero and Heidel, van der Waerden, Fraenkel and Vlastos.
" Cf. Euclid, Elnnmts X app. xxvii (Heiberg).
30
An ingenious but infertile device. A rule was devised for constructing a series of fractions
approaching as close as you please to the ratio between side and diagonal: the lines were
56 /. The Presocratics
the reaction of Eudoxus: to remodel the theory radically by allowing the
concepts of addition and greater and less to range over rationals and irrationals
alike.
This is enough to certify that the discovery of incommensurables was a real
crisis in mathematics, and to introduce another type of reaction to it. Some
mathematicians gave up the model of a line as a multiple of unit parts, a model
which made sense only on the old theory of proportion. They said instead, as
Newton said later,
31
that a line should be considered as generated not by the
summation of parts but by the fluxion or motion of a point: the extended iine
is the path of a moving thing without extension. This is said to be a relatively
late reaction,
32
but it is already under attack in Plato's Parmenides and
Aristotle's Physics; and this attack ushered in the period of Zeno's most
powerful influence on mathematics. So far, it is plain, Zeno has made no
appearance in the crisis. Some writers, hoping to find for him a directly
influential role in the mathematics of his day, have suggested that the new
picture of a line as the path of a moving point was a response not to the
156 discovery of incommensurable/ lines but to the arguments of Zeno; but this
seems incredible. No one who had been vexed by those paradoxes can have
hoped to evade them by introducing the idea of motion. In fact it is by an
adaptation of some ofZeno's arguments that Plato rejects the new picture of a
line; but Zeno himself had probably not talked of points and lines, and the
later and precise concept of a point as something with location but without
magnitude seems to have been produced to meet a difficulty that had little or
nothing to do with his work.
When Plato turns to attack this account of a point in the Parmenides, he
argues that a thing without parts cannot have a location.ll For to have a
location is to have surroundings, and this is to be in contact with something
on various sides at various points: but a thing without parts cannot have
different sides or points. This equation of location with surroundings is
standard with the Greeks: Zeno had built one paradox on it,
34
and Aristotle
was to give his own sophisticated version of it in the fourth Book of the Physics.
Until it was replaced by the method of fixing location by co-ordinates, the
formal objection to allowing a point location went unanswered. Aristotle
inherited it,
35
as he inherited the corollary argument that a point cannot be
said to move.
36
Moreover when Plato goes on to define the conditions under
described by a series of paired numbers such that always the square on the diagonal equalled
twice the square on the side plus or minus one, and the approximate sides and diagonals defined
by this construction were called the 'rational' sides and diagonals (Theon of Smyrna, 42. 10-44.
17 (Hiller), cf. Plato, Rtp. 546c and Proclus' Commentary, ii. 27 (Kroll)).
" Quadr. Curv. (1704), intro. 27.
"Sextus Empiricus, adv. matlt. X 281-2. The concept of a line that was superseded by the
fluxion-model is probably not the innocuous one compared with it by Sextus (279-80) and
Proclus (in Eucl. i 97-8).
"Parm. 138 a: part of the attack described by Aristotle in Mttaphysi&S 992a 20-2.
" Diels-Kranz, op. cit., 229 B 5.
"Physics 212b24-5.
" Parm. 138c-d; Phys. 240b8-241 a6.
3 . .{)no and the Mathematicians
57
which anything can be in contact with different things and, in particular, can
be a member of a linear series of such things, he provides both the pattern and
the terminology for Aristotle's own treatment of points and lines in the
Physics.
37
Aristotle's insistence that a line can be composed only of smaller,
indefinitely divisible lines and not of points without magnitude rests on
Plato's treatment of the point as a thing that cannot have sides or neighbours;
and it is more than likely that Plato's argument derives from Zeno's warning
that the parts of anything must have some magnitude, however small.
Now it is this same distinction between lines and points that Aristotle turns
against Zeno's remaining puzzle, the Flying Arrow; and his mishandling of
both the distinction and the puzzle is the last topic I want to discuss./
The arrow
Zeno's last paradox concerning motion is given by Aristotle in a form which,
despite the depravity of the text, can be articulated as follows: Anything
which occupies a space just as its own size is stationary. But in each moment
of its flight an arrow can only occupy a space just its own size. Hence at each
moment of its flight the arrow is not moving but stationary. But what is true
of the arrow at each moment of a period is true of it throughout the period.
Hence during the whole time of its flight the arrow is not moving but
stationary.
38
Aristotle says that the fallacy lies in assuming that any stretch of time is a
collection of moments, a mistake parallel to thinking that any line is a
collection of points. Now in a sense his diagnosis is right; but not in the sense
that he gave to it. Before we come to this, however, one small point needs to
be made. Aristotle is often represented as accusing Zeno of thinking that any
time-stretch consists of a finite collection of moments. But we shall see that
Zeno does not need this premiss (nor its denial, either). And as for Aristotle,.
he was equally anxious to deny that a period could be composed either of a
finite or of an infinite number of moments. Define moments as having no
magnitude, and Aristotle has learnt from Zeno to argue that no magnitude
can be in either of these ways a sum of such parts.
Let us clear some issues by an imaginary conversation.
Aristotle: You claim that (a) in each moment of its flight the arrow must be
stationary, since evidently it has no time to move; but (b) what is true of it at
"Terminology: conla&l (llllplesllllli), in succession (epllexis), neighbouring (echeslhai), Parm. 148e
and Phys. 226b18 ff. Plato defines the first by means of the other two, Aristotle defines the last by
the first two.
"Aristotle, Physics 239b5-9; 30-31: on the text cf. H. D.P. Lee, .(eno of Elea (Cambridge 1936),
78-81. [The use of the term 'now' to mark the durationless instant, a term on which Aristotle's
discussion of the A"ow so much relies, may not have been Zeno's. Perhaps the word is being
groomed for this role in Plato's Parmmides 152b 2-e3; perhaps on the other hand Parmenides B 8.
5 suggested its exploitation to Zeno. But even if Zeno cast his premisses in terms of hopolan and
aei, Aristotle can hardly be wrong in treating the puzzle as resolving a period of time into
constituents which (so to speak) leave no room for movement.]
157
58
/. The Presocratics
each moment is true of it throughout the whole period. Hence your conclusion.
But you agree that moments have no magnitude (that, of course, is why the
arrow cannot move in one). Consequently they cannot be added together to
make a period of time, which does have a magnitude.
-?,eno: You seem to be attacking my premiss (b). I grant what you say:
indeed my argument depends on stressing this characteristic of points and
moments. (You remember that I was accused of overlooking it last time). But
the argument does not require that the moments should be added together. I
merely assumed that if something was true at any and every moment of a
period it was true throughout a period. It is ordinary sense and not bad logic
158 to say that if at any moment this afternoon I was asleep-/ at 4:30 as well as 2,
and at any such precise time you care to take- then I was asleep throughout
the afternoon.
Aristotle: But you cannot describe periods exhaustively in this way, in terms of
moments. However many moments you can mention you are still only
specifying the limits of the periods that separate them, and at any stage of the
division you like it is these periods that make up the overall period. You can
never have two neighbouring moments. So if it is correct to infer from the fact
that at any time this afternoon I was asleep, to the fact that I was asleep all
afternoon this can only be because 'at any time' means 'at all periods,
however small'. And 'at 4:30' can only mean, in this context, 'at some period
however small round 4:30'. Don't misunderstand me: I am not suggesting
that such time-references as '4:30' are really specifications of periods oftime:
if they were, we should have to invent a new set of time-references to say
when such periods began and ended; and it is absurd to ask how long 4:30
lasts. I am only suggesting that here what parades as a time-reference must be
a shorthand specification for some small period of time .
.:(,eno: In as far as this argument differs from your first, it is trifling. To
specify moments is surely enough to specify the limits of periods. But to say
that therefore any formula phrased in terms of moments is indirectly about
periods oftime merely invites the converse reply: for to identify a period is to
describe the moments that define it.
Aristotle: Nevertheless you do talk about moments in a way that is only
appropriate to periods. You say, for instance, that the arrow is stationary at
every moment of its flight. But in the section of my Peri Kiniseos which
introduces an attack on your paradoxes
39
I show that if there is no time in a
moment for the arrow to move there is no time for it to be stationary either.
Movement involves having different positions at different moments, and
accordingly rest involves having the same position at different moments. But
we are considering only one moment, so neither of these ideas applies. In
making either of them apply you treat the single moment as a period of time
itself containing different moments.
-?,eno: Now, in effect, you are turning your attack to my premiss (a). But if it
is true that at any moment of its flight the arrow is neither moving nor at rest
" Physics VI 239a23-239b4.
3. Zeno and the Mathematicians 59
then, by my second premiss, the/ arrow is throughout its flight neither 159
moving nor at rest. And as a paradox that will do - unless you can find some
independent argument against my second premiss. Of course, if that premiss
also depended on treating moments as small periods, the argument would
collapse. But you have not shown this so far.
Aristotle: It might be shown like this. Consider a spatial analogy to your
argument about time. If a surface is uniformly red all over it is red in every
part of it, however small the part. But it is not red or any other colour at every
point, if by 'point' you mean something without extension. In the ordinary
sense of 'red' we have no use for calling something without extension red. If
we had such a use it must be because 'red' was used here in an unfamiliar
sense. Likewise, even if it were legitimate to infer from 'The arrow was
moving (or at rest, or neither) throughout the period', this could only show
that the expression 'moving' (or the expression 'at rest', or both) was being
used ambiguously between the two cases. Your second premiss, if it is true,
rests on a pun; but if it rests on a pun the conclusion you want will not follow .
.(eno (by now a prey to sharp anachronism): This is surely wrong. For
suppose a body is constantly increasing its speed: this state of affairs is
naturally explained by saying that at any moment it moves at a speed greater
than at any previous moment since its motion began. And here notice that the
verb 'to move' is associated with the common expressions for velocity and that
it can be paraphrased by the common equivalents, 'to change position' and so
forth. So it is false that, if'motion at an instant' had any use, 'motion' would
have a different sense here from that which it usually carries.
Arbiter: You are both right and both wrong. Consider again the expressions
'X was moving at some moment t', 'X was moving throughout the period p'.
Aristotle denied that the expression 'X was moving' had the same sense in
both contexts. And in face of Zeno's reply we can add that any expansion of
the expression, such as 'X was moving at velocity V', could not have the same
sense in both. For consider how the methods of confirmation differ. Velocity
is distance measured against time. The simple question, With what velocity
did X traverse din the periodp? gets the simple answer, dfp. But the question,
With what/ velocity was X moving at a time t inside the period? is complex. It 160
calls for the concept of a limit - the possibility of measuring an indefinitely
long series of distances against a corresponding series of times. It can be
answered, for instance, by constructing a graph whose curve is indefinitely
corrigible by further pairs of measurements. To be sure, once we have this
graph we can replace our simple questions about speed over a period with a
more sophisticated one. For whereas our first question merely demanded the
overall speed (not the average speed: this is again complex), we can ask now
whether X 's speed over the period was constant. And this involves a different
use of the graph. To say that X moved with a constant speed during the period
is to say something doubly general, when to ascribe it that speed at one
moment is to say something singly.general: for now we ascribe it a speed at
each moment in the period. But the possibility of operating on either of these
levels of generality depends on being able to answer questions of our first,
simple form, and the converse is not true. And thus Zeno's rejoinder fails. For
60
/. The Presocratics
since in this way the possibility of talking about motion at a moment rests on
the possibility of talking of motion over a period, the two uses of' motion' are
not the same. Likewise we could if we wished give a use to the expression
'colour at a point' by building on our ways of describing a colour over a space,
but we could not begin the other way round without a radical change in the
use of colour-words. But in another way Zeno was right. For to say that these
are not the same use is not at all to say that Zeno's second premiss depends on
a pun. The premiss is valid, and it is valid precisely because it is the sort of rule
whereby we do give a use to such an expression as 'moving at a moment'. We
rule that, when and only when it is correct to say 'X was moving throughout
the period p', it is also correct to say 'X was moving at any moment tin p'.
Aristotle's fallacy lay in supposing that to infer from the second formula to the
first, one must regard the second as specifying a conjunction of moments
exactly as long as the period specified in the first. He was in fact applying a
simple model of induction, that model which set a premium on the exhaustive
enumeration of cases and which Aristotle took to require strict synonymy
between different occurrences of the predicate ('X-moving', for instance, in
the inference from 'Each moment inp is a case ofX-moving' to 'pis a case of
161 X-moving'). And thus he failed to grasp that the two/ senses of'moving' are
not identical but yet systematically connected; and his failure to see this
connection between the common uses of a common word led him to rule out
one use entirely in favour of the other. His reply to Zeno rejects all uses of
'movement' other than that which can be described in terms of periods of
time, just as the colour-model we considered exhibited all uses of 'red' as
applicable to colour-stretches. And this is an unjustified departure from
usage: it deprives us of a convenient method of characterizing motion which is
common idiom for us and for the Greeks.
Now (and here we can drop the pretence of dialogue) if this is so Zeno's
fallacy cannot lie in his second premiss. Therefore it lies in premiss (a), and in
particular in the proposition 'There is no time to move in a moment' (with or
Aristotle's rider: 'and no time to rest either'). The picture we are
g1ven 1s of the arrow bottled up in a piece of time that fits it too closely to allow
any movement. The moment is too short to fly in. But such talk of movement
is appropriate only when we have in mind periods of time within which
movements could be achieved. It is not false that movements can be achieved
w_ithin moments: it is absurd either to say or to deny this, for moments are not
of time such that within them any process can either take place or lack
the t1me to take place. But this certainly does not show that the arrow is not
moving at any moment. It is, of course: we have seen the sense in which it is.
Whether it is, is a question of fact and not of logic.
So, despite his contrast between moments and periods of time, Zeno was
treating moments as stiflingly small periods. To that extent Aristotle was
in ?is diagnosis. But he did not apply the diagnosis where it was needed.
H_1s that t_here. can be any talk of motion except in direct connection
of time 1s a surr_ender to Zeno; and his failure to come to grips
With premiss (a) compels h1m to struggle against the wholly respectable
premiss (b).
3. Zeno and the Mathematicians 61
This surrender to Zeno had notable results in the history of dynamics.
Notoriously, Aristotelian dynamics failed to deal adequately with
acceleration; and it might be thought from what has been said that the
failure lay in insisting that acceleration (a phenomenon which Aristotle
certainly took seriously) must be analysed in terms of motion and speeds over
periods of time, andjnot in the more manageable shorthand of velocity at an 162
instant. But this is not the root-issue. Unable to talk of s p ~ e d at an instant,
Aristotle has no room in his system for any such concept as that of initial
velocity, or what is equally important, of the force required to start a body
moving. Since he cannot recognize a moment in which the body first moves,
his idea of force is restricted to the causing of motions that are completed in a
given period of time. And, since he cannot consider any motion as caused by
an initial application of force, he does not entertain the Newtonian corollary of
this, that if some force F is sufficient to start a motion the continued
application ofF must produce not just the continuance of the motion but a
constant 'change in it, namely acceleration.
40
It is the clumsy tools of
Aristotelian dynamics, if I am right, that mark Zeno's major influence on the
mathematics of science.
40
He would have had another reason for rejecting Newton's account of acceleration, for that
account holds good only in a vacuum, and Aristotle thought a vacuum impossible. But some of
his followers re-imported the vacuum without abandoning the rest of the system.